Space Human-Robot Interaction with Gaze Tracking Based on Attention Mechanism.

Area of Science:

• Robotics
• Human-Computer Interaction
• Aerospace Engineering

Background:

• Gaze is a natural non-verbal communication method ideal for human-robot interaction.
• Current space-based human-robot interaction using gaze is underdeveloped.
• The AAR-2 free-flying astronaut-assistant robot platform is utilized.

Purpose of the Study:

• To explore and establish gaze-based human-robot interaction for space environments.
• To develop an accurate and real-time gaze tracking system for astronaut-robot collaboration.
• To enable intuitive control of the AAR-2 robot through gaze direction.

Main Methods:

• Established a gaze tracking database using the AAR-2 robot.
• Constructed an attention mechanism-based gaze tracking model.
• Developed a human-robot interface for gaze command transmission via radio frequency.
• Implemented an efficient image preprocessing method for real-time performance.

Main Results:

• Achieved an effective gaze tracking model with high accuracy.
• Implemented an efficient image preprocessing method with high real-time performance.
• Demonstrated that the system meets practical requirements for accuracy and real-time processing in space applications.

Conclusions:

• The developed system enables effective gaze-based human-robot interaction in space.
• The AAR-2 robot can be controlled intuitively via gaze, with operational feedback.
• The system shows promise for enhancing astronaut efficiency and safety in space missions.